On the Distortion of Cirrus Logic CS431xx-Based Devices: A Comparative Review

[carlcamper]

Sad to say that my Moondrop x Crinacle Dusk DSP cable exhibits the Cirrus Hump distortion. However, my Moondrop Quark 2 with its fixed USB-C DSP doesn't!

 

[CedarX]

Sad to say that my Moondrop x Crinacle Dusk DSP cable exhibits the Cirrus Hump distortion. However, my Moondrop Quark 2 with its fixed USB-C DSP doesn't!

Could you show the actual measurements?

I don’t have Dusk, but I believe it uses the same Conexant CX31988 chip as the FreeDSP cable; Quark 2 uses a TTGK/CBHT CB1200AU chip. The Dusk actual measurements are very interesting: either it is Cirrus-based and not Conexant, or the CX31988 has something similar to Cirrus’ DRE.

 

[carlcamper]

Could you show the actual measurements?

I don’t have Dusk, but I believe it uses the same Conexant CX31988 chip as the FreeDSP cable; Quark 2 uses a TTGK/CBHT CB1200AU chip. The Dusk actual measurements are very interesting: either it is Cirrus-based and not Conexant, or the CX31988 has something similar to Cirrus’ DRE.

Sorry, I only tested with the Cmaj file, so no visual measurements.

 

[reallystate]

Sorry, I only tested with the Cmaj file, so no visual measurements.

Can you record it via Audacity?

 

[IsaacOscar]

Can you record it via Audacity?

Unless they have a 0.78mm 2pin lin-in port or appropriate adapter I don't think that will work...
(maybe some soldering would work...)

 

[CedarX]

Unless they have a 0.78mm 2pin lin-in port or appropriate adapter I don't think that will work...
(maybe some soldering would work...)

The Yongse Y01 dongle reviewed by Amir in May-2024 uses a Conexant (Synaptics) CX31988, same as Moondrop FreeDSP cable (and Dusk?).
Although the FW may be different, this would be an easier way to measure the presence of a “Cirrus hump” and DRE-like feature on this chip.
At the same time, the CX31988 is not very popular… May not be worth the measurement efforts. Its little brother, the CX31993, is more interesting IMO: much more common.

 

[MBO]

I need to ask somehow if Cirrus is going to somehow update their 43198 series and fix their bugs.

 

[tccalvin]

I need to ask somehow if Cirrus is going to somehow update their 43198 series and fix their bugs.

Who knows...

But the chips are user-programmable so the issue can be fixed by the manufacturers.

 

[CedarX]

I need to ask somehow if Cirrus is going to somehow update their 43198 series and fix their bugs.

Who knows...

But the chips are user-programmable so the issue can be fixed by the manufacturers.

@tccalvin is correct, this is not a bug, and the DRE feature in itself is not necessary a problem. We've now learned it can be disabled or optimized—some dongles let you turn DRE on/off, or set to "optimum" parameters. Cirrus was probably greedy, skewing the default DRE settings toward best measured performances at the detriment of audible artifacts. With DRE disabled, the CS431xx may not perform much better that their old CS4398 DAC chip (~2002 vintage).

As with the ESS hump, these companies should do a better job communicating to their OEM's how to best implement their chips and what are the potential tradeoffs: that's what Cirrus needs to fix. The OEMs also need to pay much more attention and do more thorough testing. Thankfully, ASR members are here to catch these issues !!!

 

[MaxwellsEq]

Cirrus was probably greedy, skewing the default DRE settings toward best measured performances at the detriment of audible artifacts

This feels correct to me. Cirrus are in a bit of a bind though. If they advise manufacturers to set DRE to off or least-impact then "classic" noise and distortion measurements and SINAD numbers will appear less stellar. The measurements will still be sufficient, of course.

Which also unfortunately still leaves people with devices where firmware can't be changed out in the cold having to live with the DRE clicks and hump.

 

[Sokel]

The need to show competitive measured performance probably cuts both ways, be it DRE, enormous voltage outputs to match AP's sweet spot, etc.

Engineers...

 

[scruffy1]

hahaha, as a medico i always do a double take when i spy DRE

 

[confucius_zero]

No cs-hump for the ka11?

 

[L5730]

Am I right in thinking the KA11 has it's own problem in terms of heat during use?

 

[tccalvin]

No cs-hump for the ka11?

It's not possible to tell from this test. I believe SMPTE/DIN Ratio only consists of two tones (useful to detect IMD), while the test needed to detect the hump is a 32-tone multitone level sweep.

I know the descending line looks similar between the two tests, but we should pay attention to what's written on the graph as well.

With all likelihood, the KA11 exhibits DRE artifacts just like every other CS431XX DAC.

I have requested that reviews include the 32-tone sweep test and/or the CMaj test, but the request was denied. Another user had requested a different test. Denied. So, as far as Amir's reviews are concerned, we'll likely never be able to tell wether or not a DAC has DRE issues.

 

[SoySalsaPicante]

...So, as far as Amir's reviews are concerned, we'll likely never be able to tell wether or not a DAC has DRE issues.

One surefire way to know for a fact that a reviewed device does not have Cirrus Logic DRE issues is to immediately disregard the product if it uses a Cirrus Logic chip DAC.

The only way to teach them a lesson is to boycott their stuff, and stick to ESS, AKM, Rohm, etc.

 

[tccalvin]

One surefire way to know for a fact that a reviewed device does not have Cirrus Logic DRE issues is to immediately disregard the product if it uses a Cirrus Logic chip DAC.

The only way to teach them a lesson is to boycott their stuff, and stick to ESS, AKM, Rohm, etc.

Yes... Obviously it doesn't help if Amir keeps omitting relevant measurements and giving perfect scores to CS DACs. But he has made it clear that he doesn't care.

 

[danadam]

It's not possible to tell from this test. I believe SMPTE/DIN Ratio only consists of two tones (useful to detect IMD), while the test needed to detect the hump is a 32-tone multitone level sweep.

It doesn't necessarily have to be a multitone. The first post shows TDFD Bass test, which is 41 Hz + 89 Hz 1:1, and it also shows the hump.

Here's TDFD Bass (41 Hz + 89 Hz, 1:1) vs SMPTE (60 Hz + 7 kHz, 4:1) for Tanchjim Space:

 

[MBO]

Luxsin replied to me, confirming that the new DAC X8 has an exclusive switch for the DRE function.

 

[zany.ninja]

Just a little observation as I don't think anyone has commented on this.

In the screenshots of @jkim 's tests in the first post of this thread, the JCally JM20 Max has ultrasonic images/reflections of the playback signal (test signal in this case), highlighted in red below:

While the JCally JM20/JM28 is clean:

Does the JM20 Max do this at all times? If yes, what could be causing it? The SGM8262-2 in the JM20 Max? Or something else?